Paralleling the exponential growth in computing muscle is an explosion in demand for communications bandwidth. Faster, cheaper access to ever-increasing volumes of data will be a major force in shaping the way we use computer technology and the way in which that technology evolves. Optical-fiber links are a key element of this bandwidth explosion.
They have enabled the adoption of telecommunications standards with bit rates far higher than were possible using copper cables. Advances in fibers, lasers and photodetectors have allowed the introduction of telecom standards with data rates higher than 10 Gbits/second. The adoption of dense wavelength-division multiplexing permits many independent data channels carried on different optical wavelengths to coexist on one fiber. This has further increased available network bandwidth. Meanwhile, an increase in the variety and volume of data traffic passing through these fiber links brings new challenges as service providers manage and expand their networks.
Bottleneck at switch
One potential technology bottleneck is that of telecom switches, the essential network elements that allow traffic to be routed from a source to a destination. An optical cross-connect switch may be thought of as a black box with multiple input and output fibers carrying network traffic. The basic function of the switch is to allow the signal on any one of the input fibers to be redirected to any one of the output fibers in the manner configured by the users.
The optical cross-connect switches used in today's networks rely on electronic cores. An optical signal arriving at a switch input port is converted to an electronic signal by a high-speed photodetector (receiver). Electronic circuits in the switch core then direct the signal to the desired output port. A final electrical-to-optical conversion is performed by a laser diode, transforming the signal back into light for onward transmission in the fiber network.
The fundamental problem with these electronic cores is that they do not scale well to large port counts (numbers of input and output channels) and are costly to replace for network upgrade to higher data rates needed for the growing demand for bandwidth. Many in the industry now believe that electronic-core cross-connect switches cannot efficiently meet the needs of tomorrow's communication networks, so a major challenge is to develop new, all-optical switching technologies that can fill this gap. To gain acceptance, these new technologies must be able to demonstrate low-optical-loss switching with extremely high reliability to meet the high-availability requirement of the communication service providers.
Among the candidates competing to replace electronic cores are optical switches based on thermal bubble generation in planar waveguides (developed by Agilent Technologies) and those using the electro-optic properties of liquid crystals (under development by Chorum, Corning and Spectra-Switch). Although some of these products are closer to commercialization in the low-port-count (two to 32 channels) switch configurations, it remains uncertain whether they can be extended beyond their current forms. In addition to the relatively high optical throughput losses of those devices, the need for temperature control and uncertainties over long-term aging effects are common causes of concern.
A number of companies including Nortel Networks, Lucent Technologies and Onix Microsystems are backing silicon micromachining (known as microelectromechanical systems, or MEMS) technologies to emerge as the leader. Silicon MEMS technology uses processing steps derived from the IC fabrication techniques of photolithography, material deposition and chemical etching to produce movable mechanical structures on a silicon chip.
Often with integrated microactuators built into them, these MEMS devices are typically addressed using electrical signals to produce controlled motion of the micromechanical structures on the chip. The batch-processed nature of the fabrication procedure means that MEMS devices can be mass-produced on a large scale, which reduces manufacturing costs. In addition, the extremely small physical sizes and masses of these silicon "machine parts" often make them more robust and capable of faster operation than conventional macroscopic mechanical devices.
With the close affiliation of its founding members with the University of California at Berkeley (one of the world's leading centers for MEMS research), Onix Microsystems is bringing optical MEMS to the market with a range of optical switching products based on patented micromirror technologies. The key element of its switching engines is a two-dimensional matrix of micromachined mirrors, fabricated in single-crystal silicon, that is used to make reconfigurable optical connections between input and output fiber arrays. One particular Onix approach uses a two-dimensional N x N mirror matrix combined with linear input and output fiber arrays to form an N x N crossbar switch. Control signals applied to the MEMS chip fix the position of each individual mirror to either pass or intersect the input light beams, directing each incoming light signal to the desired output port.
Because the intersecting micromirrors must direct the outgoing light beams into small-diameter fiber cores (typically fewer than 9 microns across), tight control of the mirror angles is vital to minimizing the optical power loss incurred from passing through the switch. Extremely low insertion loss and channel crosstalk can be achieved by this architecture to produce compact, all-optical switch modules with up to 64 input/output ports.
These relatively small port-count switches are useful, for example, for network protection and restoration-to allow carriers to quickly reroute data traffic around fault locations in the event of fiber breaks or equipment failures. However, central offices often require larger switches with hundreds and up to thousands of input/output ports. To scale up the channel count beyond 32 x 32 ports, two solutions are available depending on the specific requirements of the end system.
One is to use the smaller switch modules as building blocks, cascading them to form a larger switch. A three-stage network configuration allows one to construct a fully nonblocking cross connect with up to 1,024 x 1,024 channels by linking together 32 x 32 switch modules. Alternatively, instead of using N2 mirrors-each with two controlled positions-to form an N x N switch, one can use 2N mirrors each having N controllable positions to achieve the same functionality. This latter architecture, called the "steered beam" configuration (also often referred to as the "analog" or "3-D" approach), is used to construct large optical cross-connect switches within a single switch stage. The micromirrors that direct the free-space light paths are even more closely controlled by servo-feedback loop electronics and therefore give a lower insertion loss than the N2 mirror configuration is capable of.
Switching is accomplished by manipulating the free-space propagation paths of the light beams directly. As a result, the MEMS switches are optically transparent. In other words, they can be used in networks with widely varying data rates, modulation formats and signal wavelengths. Moreover, changes in the network properties have no effect on the switch functions. Therefore, no hardware modification is needed when the communication service providers upgrade the other parts of their networks or systems.
This last feature will increase in importance as high-bandwidth connections become more popular with consumers. With the large variety of data traffic passing through the ever more complicated communication links from local-area to global networks, the ability to work with previously installed equipment while accommodating future upgrades will be the key feature that makes all-optical switches the preferred choice of carriers.
As high-speed data networks further penetrate the consumer market, network service providers are turning to technology solutions that will enable them to offer diverse services with varying bandwidth demands. The vision of a future all-optical network that can deliver these potentials seems less and less like a distant dream. All-optical switches and MEMS-based optical switches in particular are going to be the key enablers for the future of ultrahigh-bandwidth, all-optical communication networks.